def main(author): print('OpenCV version {} '.format(cv2.__version__)) current_dir = os.path.dirname(__file__) training_folder = os.path.join(current_dir, 'data/training/', author) test_folder = os.path.join(current_dir, 'data/test/', author) training_data = [] training_labels = [] for filename in os.listdir(training_folder): img = cv2.imread(os.path.join(training_folder, filename), 0) if img is not None: data = preprocessor.prepare(img) training_data.append(data) training_labels.append([0, 1] if "genuine" in filename else [1, 0]) test_data = [] test_labels = [] for filename in os.listdir(test_folder): img = cv2.imread(os.path.join(test_folder, filename), 0) if img is not None: data = preprocessor.prepare(img) test_data.append(data) test_labels.append([0, 1] if "genuine" in filename else [1, 0]) return sgd(training_data, training_labels, test_data, test_labels)
def main(): print('OpenCV version {} '.format(cv2.__version__)) current_dir = os.path.dirname(__file__) author = '021' training_folder = os.path.join(current_dir, 'data/training/', author) test_folder = os.path.join(current_dir, 'data/test/', author) training_data = [] for filename in os.listdir(training_folder): img = cv2.imread(os.path.join(training_folder, filename), 0) if img is not None: data = np.array(preprocessor.prepare(img)) data = np.reshape(data, (901, 1)) result = [[0], [1]] if "genuine" in filename else [[1], [0]] result = np.array(result) result = np.reshape(result, (2, 1)) training_data.append((data, result)) test_data = [] for filename in os.listdir(test_folder): img = cv2.imread(os.path.join(test_folder, filename), 0) if img is not None: data = np.array(preprocessor.prepare(img)) data = np.reshape(data, (901, 1)) result = 1 if "genuine" in filename else 0 test_data.append((data, result)) net = network.NeuralNetwork([901, 500, 500, 2]) net.sgd(training_data, 10, 50, 0.01, test_data)
def run(args): text = args['text'] or main.read_whole_file(args['file']) p_params, l_params = main.separate_params(args) prep_text = preprocessor.prepare(text, args['mode'], **p_params) min_d = (args['min_dictionary_absolute'] or 0) / len(prep_text) or args['min_dictionary_relative'] max_d = (args['max_dictionary_absolute'] or 0) / len(prep_text) or args['max_dictionary_relative'] freq = stats.relative_frequency(prep_text, args['ngram']) vocab = [k for k, v in freq.items() if v >= min_d and v <= max_d] res = [] output_file = args['output_file'] for v in vocab: args['template_string'] = v args['output_file'] = None args['aprox_output_file'] = None _, gamma = main.run(args, False) res.append((v, gamma)) if output_file is not None and len(output_file) > 0: main.write_to_file(output_file, res, False)
def run(args, visualize=True): # print(args) text = args['text'] or read_whole_file(args['file']) p_params, l_params = separate_params(args) prep_text = preprocessor.prepare(text, args['mode'], **p_params) print('text legth: ',len(prep_text)) print() encoded_text = prep_text if args['mode'] != 'prep': if args['template_string'] is None: encoded_text = fluctuation.encode_vocab(prep_text) else: template = ( args['template_string'] if args['case_sensitive'] else args['template_string'].lower() ).split() encoded_text = fluctuation.encode(prep_text, template) if args['only_encode']: if args['output_file']: write_to_file(args['output_file'], encoded_text, True) return encoded_text, None result, gamma = analyse(encoded_text, args['mode'], l_params, visualize) if args['output_file']: write_to_file(args['output_file'], result, False) if args['aprox_output_file']: write_to_file(args['aprox_output_file'], [gamma], False) return result, gamma
#This takes the center of the stack as a center frame(s) center = np.int(n_total_frames) // 2 #Check this in case we are in double exposure if center % 2 == 1: center -= 1 if metadata["double_exposure"]: metadata["double_exp_time_ratio"] = metadata["dwell1"] // metadata[ "dwell2"] # time ratio between long and short exposure center_frames = np.array([raw_frames[center], raw_frames[center + 1]]) else: center_frames = raw_frames[center] #print('energy (eV)',metadata['energy']) metadata, background_avg = preprocessor.prepare(metadata, center_frames, dark_frames) #print('energy (J)',metadata['energy']) #we take the center of mass from rank 0 #metadata["center_of_mass"] = mpi_Bcast(metadata["center_of_mass"], metadata["center_of_mass"], 0, mode = "cpu") io = ptycommon.IO() output_filename = os.path.splitext(json_file)[:-1][0][:-4] + "cosmic2.cxi" if rank == 0: #output_filename = os.path.splitext(json_file)[:-1][0] + "_cosmic2.cxi" printv("\nSaving cxi file metadata: " + output_filename + "\n") #data_dictionary["data"] = np.concatenate(data_dictionary["data"], axis=0)